1. Field of the Invention
The present invention relates generally to an improved data processing system, and in particular, to a computer implemented method for managing system performance in a data processing environment. Still more particularly, the present invention relates to a computer implemented method, system, and computer usable program code for improving processor performance during power supply failure.
2. Description of the Related Art
A data processing system may experience a loss of electrical power during the data processing system's operation. For example, a power supply unit supplying electrical power to the data processing system may fail to supply electrical power to the data processing system. A power supply unit is a component of a data processing system that transforms, converts, or otherwise conditions electrical power received from the power grid and provides the transformed, converted, or conditioned electrical power to one or more other components in the data processing system.
In many data processing systems, more than one power supply units supply power to the various components of the data processing system. Multiple power supply units are often used to provide redundancy so that a catastrophic failure and complete shutdown of the data processing system can be avoided. Sometimes, several power supply units are used to enable the operation of the data processing system under partial power conditions. More than one power supply units are common in data processing systems, and especially in data processing systems with several processors.
When a power supply unit fails to provide electrical power as expected, the data processing system is said to be experiencing a power supply failure. In some data processing systems, the system must continue operating even when a power supply fails. To continue operation under reduced power due to a power supply failure condition, some data processing systems shut down or reduce power supply to certain components within the data processing system.
Generally, a processor in a data processing system is one of the largest consumers of electrical power. In case of a power supply failure, many data processing systems employ techniques to reduce the power consumption of one or more processors in those data processing systems.
One common technique to reduce a processor's power consumption is called processor throttling. Processor throttling is the process of inserting no-operation instructions in the processor's instruction queue. A no-operation instruction is an instruction that instructs the processor to not perform any operation. Processor throttling using no-operation instructions effectively idles the processor for a significant number of processor cycles, in effect dropping the power consumption of the processor.
The data processing system's overall performance deteriorates significantly when a processor is throttled. In many cases, processor throttling can cause the performance of a data processing system to drop by as much as eighty-five percent of pre-throttling performance. Thus, while the data processing system may continue to operate under a power supply failure condition, the drop in performance from processor throttling can limit productive use of the data processing system.